摘要:
Herein disclosed is a sound signal processing apparatus (10), comprising: a speaker unit (12) for converting a first sound signal to a first sound; sound signal producing means (13) for producing a second sound signal constituted by at least two different components including an echo component indicative of the sound outputted by the speaker unit (12), and a voice component indicative of one's voice having a least one leading end; echo component suppressing means (14) for suppressing the echo component of the second sound signal on the basis of the first and second sound signals to output, as a third sound signal, the suppressed second sound signal; sound signal storing means (15) for storing the third sound signal outputted by the echo component suppressing means (14); voice detecting means (16) for detecting the leading end of the speaker's voice on the basis of the third sound signal outputted by the echo component suppressing means (14); and controlling means (17) for controlling the sound signal storing means (15) to have the sound signal storing means (15) output, as a fourth sound signal, said third sound signal stored in the time period when said voice is detected on the basis of said third sound signal outputted by said echo component suppressing means, the controlling means (17) being operative to specify two different clock times on the basis of a predetermined time difference, the clock times including a first clock time at which the leading end of the voice is detected by the voice detecting means (16), and a second clock time prior to the first clock time, the controlling means (17) being operative to have the sound signal storing means (15) start to output the third sound signal stored after the second clock time.
摘要:
A noise robustness method operates jointly in a signal domain and a model domain. For example, energy is added in the signal domain for frequency bands where an actual noise level of an incoming signal is lower than a noise level used to train models, thus obtaining a compensated signal. Also, energy is added in the model domain for frequency bands where noise level of the incoming signal or the compensated signal is higher than the noise level used to train the models. Moreover, energy is never removed, thereby avoiding problems of higher sensitivity of energy removal to estimation errors.
摘要:
An image display panel includes: an image display unit including pixels arranged in horizontal and vertical directions, and displaying a composite image including different images; and an image separating unit facing the image display unit, and separating the composite image into the different images using light-transmitting portions transmitting light and light-shielding portions shielding the light. Each of the pixels includes sub-pixels each including a light emitting region, the sub-pixels have sizes different from sizes obtained by equally dividing the pixel, the light-transmitting portions are open in a diagonal direction, and each of boundary lines between the light-transmitting portions and the light-shielding portions is determined by moving parallel a continuous line that connects lines each connecting gravity centers of adjacent sub-pixels or adjacent light emitting regions in the sub-pixels, or to be a continuous line that connects lines each connecting midpoints of the gravity centers.
摘要:
Model compression is combined with model compensation. Model compression is needed in embedded ASR to reduce the size and the computational complexity of compressed models. Model-compensation is used to adapt in real-time to changing noise environments. The present invention allows for the design of smaller ASR engines (memory consumption reduced to up to one-sixth) with reduced impact on recognition accuracy and/or robustness to noises.
摘要:
A noise robustness method operates jointly in a signal domain and a model domain. For example, energy is added in the signal domain for frequency bands where an actual noise level of an incoming signal is lower than a noise level used to train models, thus obtaining a compensated signal. Also, energy is added in the model domain for frequency bands where noise level of the incoming signal or the compensated signal is higher than the noise level used to train the models. Moreover, energy is never removed, thereby avoiding problems of higher sensitivity of energy removal to estimation errors.
摘要:
An image display device includes an image display, an image separator and a controller. The image display displays a plurality of parallax images. The image separator performs optical separation such that the plurality of parallax images which are displayed are viewed as a stereoscopic image. The image separator includes a light-transmitting region (region 0), a light-masking region (region 1) and an intermediate region (region 2) which is located between the light-transmitting region (region 0) and the light-masking region (region 1). The controller controls light-transmittance of the intermediate region (region 2).
摘要:
A video processing device is provided, in which the cross-talk is prevented from increasing while the Moiré pattern is diminished. A video display device displays multiple-parallax video using a parallax barrier method, the video display device comprising: a display unit displaying pixel columns of a given image and pixel columns of one or more parallax images corresponding to the given image, all of the pixel columns being arranged in predetermined order; and a parallax barrier arranged at a predetermined distance from an image display face of the display unit, and having a plurality of elongated passing regions passing light rays from the display unit, wherein at least one of the passing regions has a waveform formed at a longitudinal edge thereof, and at least one waveform includes two wave units that differ in terms of wave height or wave width.
摘要:
A video processing device is provided, in which the cross-talk is prevented from increasing while the Moiré pattern is diminished. A video display device displays multiple-parallax video using a parallax barrier method, the video display device comprising: a display unit displaying pixel columns of a given image and pixel columns of one or more parallax images corresponding to the given image, all of the pixel columns being arranged in predetermined order; and a parallax barrier arranged at a predetermined distance from an image display face of the display unit, and having a plurality of elongated passing regions passing light rays from the display unit, wherein at least one of the passing regions has a waveform formed at a longitudinal edge thereof, and at least one waveform includes two wave units that differ in terms of wave height or wave width.
摘要:
An image display device includes: an image display unit (100) that displays a plurality of parallax images corresponding to the number of view points for stereoscopic viewing on a screen of a display panel (100a); a parallax barrier (101) that is arranged facing the screen, and performs optical separation of the displayed parallax images; and an adaptive pitch adjustment unit (103) that adjusts, for each of regions of the display panel (100a), perform adjustment on a pitch of the parallax barrier (101) corresponding to the region in accordance with an amount of distortion of the display panel (100a) in the region.
摘要:
An image display device includes an image display, an image separator and a controller. The image display displays a plurality of parallax images. The image separator performs optical separation such that the plurality of parallax images which are displayed are viewed as a stereoscopic image. The image separator includes a light-transmitting region (region 0), a light-masking region (region 1) and an intermediate region (region 2) which is located between the light-transmitting region (region 0) and the light-masking region (region 1). The controller controls light-transmittance of the intermediate region (region 2).